The Fundamental Review of the Trading Book is a particularly challenging piece of text to implement, but banks trying to do so with the less capital-intensive Internal Model Approach are finding it particularly difficult as they face having to keep their models approved. Some are considering not bothering with it. By Farah Khalique.

Banks are inundated with new post-financial regulations, but there is one that stands out as particularly challenging: the Fundamental Review of the Trading Book (FRTB). The global financial crisis prompted financial regulators to inspect banks’ trading books, and they did not like what they saw.

“After the financial crisis, regulators were quite busy in filling the gaps they discovered. They were aware something needed to be done,” says Andreas Bohn, associate director in risk management at the Boston Consulting Group. 

Piecemeal measures were introduced, including Stressed Value-at-Risk – designed to measure potential losses during bad market conditions – but ultimately the decision was made to clean out banks’ trading books from top to bottom.

The Bank for International Settlements’ Basel Committee on Banking Supervision (BCBS) set about ensuring that banks are robust enough to survive volatile periods and ultimately avoid government bailouts.

“Regulators have in the past argued that FRTB isn’t designed to increase capital per se, but it does,” says Daniel Mayer, senior manager at Deloitte.

The go-live deadline for FRTB is December 31, 2019, although Australia’s financial regulator has told the country’s banks it will not confirm the new market standards until 2020 at the earliest. US banks are still waiting on the Fed for additional guidance.

New rules

The rationale for FRTB is that it addresses the shortcomings of the current market risk capital regulation, Basel 2.5.

First, it defines the blurred boundary between a bank’s trading book and its banking book, the latter of which has lower capital requirements. Regulators want to prevent traders from moving trades from the trading book into the banking book, to avoid higher capital charges.

FRTB also aims to fill in any gaps uncovered over the past decade in risk architectures: banks are using it as an opportunity to overhaul their market risk systems.

One of the biggest modifications is changing the fundamental methodology behind how banks calculate market risk. Previously, Value-at-Risk (VaR) was the industry standard, but this was heavily criticised in the wake of the financial crisis. Under VaR, banks did not anticipate the looming credit crash and hence were ill-prepared. The new standard is called the Expected Shortfall, which is better at measuring the probability of events or outcomes that have only a very small chance of happening, but could prove to be catastrophic.

Banks now have two approaches to calculating market risk: an individually tailored Internal Model Approach (IMA) that captures the nuances of a specific trading book, or a broader Standardised Approach (SA).

Even the Standardised Approach under FRTB is radically different from Basel 2.5, in that the capital charge is now based on sensitivities that have to be computed using front office pricing models, says Vinod Bhaskaran, global solutions lead in the trading and risk business at financial software provider Misys.

“The purpose of this is clear: to make the standardised capital charge more risk-based and aligned with the front office,” he says.

The SA needs to be available for all trading desks anyway, but banks with complex trading books may also pursue the IMA. It has the benefit of incurring fewer capital charges, but is trickier to calculate and secure the necessary regulatory approval.

Misys’ clients are seeing that FRTB SA results in a two to six times’ higher capital charge, says Mr Bhaskaran, whereas FRTB IMA results in a 1.1 to 1.5 times’ impact, a significantly lower capital charge.

Low pass rate

The IMA is where banks are struggling, say market experts. The IMA under FRTB is substantially different to that under Basel 2.5: approval is granted to individual trading desks as opposed to the entire bank’s trading floor, and is now based on passing difficult profit and loss (P&L) attribution tests and back testing.

“The P&L attribution test in particular ensures that there is a tight alignment between the front office and risk pricing models,” says Mr Bhaskaran

But if a desk fails either of its two-monthly P&L attribution tests four times or more within a year, it will lose its hard-won internal model approval and be forced to adopt a standardised model. Most banks that are testing out FRTB are experiencing high failure rates, of anywhere between 60% and 79%, by market estimates.

“That is definitely a challenge. [But] the banks are fairly sure they will have a solution in place at the time of go live. The question is what is their ambition level? How many of their desks will be on the IMA?” asks Anke Raufuss, partner at McKinsey & Company

The decision to pursue IMA or not is a question that some Tier 2 banks are grappling with, says Ted Rees, senior director at Synechron Business Consulting.

“Is it worth the investment and efforts? A few of our clients are facing a tough decision on that,” he says. “There is some pressure from regulators if you’re a globally significant bank to be mainly on an IMA. If you’re not, it’s going to be a question of what the benefits are.”

Banks are therefore assessing whether or not to invest in building huge risk models for products, or even entire desks, that may not be profitable a few years down the line. FRTB is forcing banks to consider whether they need to exit certain businesses altogether, if they are not worth the resultant capital charges or the hassle of devising an IMA to avoid such high charges.

Those further down the food chain are looking to shun the IMA in favour of a more straightforward SA. The capital charges are higher, but their trading books are smaller than those of their larger competitors.

Steve Wilcockson is a financial risk specialist from MathWorks which provides the mathematical modelling tools used by banks, asset managers and central banks. “We have not seen complete professional implementations of FRTB-ready systems yet,” he says.

Data dilemma

Banks have trouble sourcing good-quality data for risk models, says Mr Wilcockson, particularly for complex derivatives where internally-sourced granular data is limited and third-party market data sources have historically been lacking.

Charlie Browne is head of the market data and risk solutions division at GoldenSource, a data provider to banks including Deutsche Bank, HSBC and Nordea.

“From a data perspective, you need a lot more of it, and the rules that you need to do the calculations are a lot more prescriptive. Banks spent the past 10 years aggregating data not at trade level, [but rather at a] more summarised level,” he says.

The availability of public trade data varies wildly across jurisdictions and products. The US has a central reporting depository, the Trade Reporting and Compliance Engine, while the introduction in Europe next year of the Markets in Financial Instruments Directive II (MiFID II) also promises to open the gateway to more publicly available trade data. The same cannot be said of Asia-Pacific, which has few public data facilities and does not subscribe to the principles of MiFID II.

To overcome this problem, Bloomberg is working with a group of banks to pool trade data, says global head of regulatory products and reference data Chris Casey.

FRTB requires banks to calculate data at individual trade level in a specific way. To be able to model a risk factor, the bank must look at historical real trades or committed quotes, defined as at least 24 observable real prices over a rolling 12-month period with a maximum period of one month between any two consecutive observations.

Risk factors that can not be modelled in the above format are classified as non-modellable risk factors (NMRF), and incur higher capital charges.

The industry understands the merits of NMRFs, says Mr Casey, but it is tough to comply with on an ongoing basis. The cyclical nature of some asset classes means a risk factor could trade every month and comply with FRTB, but one month encounter 35 days between two trades. That instrument would then be classified as an NMRF.

“Banks are now speaking to regulators about the practicalities of complying with this. They are aware that nuanced technical aspects of requirements can be extremely burdensome,” says Mr Casey.

Operational challenges

FRTB compliance incurs huge operational challenges too, particularly adherence to the Internal Model Approach. It imposes a greater computational burden than the existing standard, management of more than 10 years of market data, calibration of a bank’s worst stress periods, scenario generation across multiple trading desks and orchestration with SA workflow elsewhere in the bank. Some desks within a single bank may choose to pursue an IMA, while others will stick to the SA.

Banks will also have to show a strong correlation between the data and calculations in their risk department and those in their finance department. Historically, these have been separate functions, says Synechron’s Mr Rees. “Risk is more forward-looking while finance is about catching up with month-end and reconciling data,” he says. “Bringing the two together is a big challenge. It is rare that banks have a single data warehouse for both risk and finance. These units have often developed along their own lines.”

Furthermore, once a bank has secured its regulatory approval for an IMA, it has to maintain it. A failure to pass the P&L attribution and back-testing tests means the desk drops down to an SA, which could lead to higher capital charges.

“It is also important to note that FRTB SA requires new prescribed sensitivities to be computed in the front-office systems and then integrated with the FRTB aggregation element, which can be a challenge for large banks with multiple front-office systems,” says Mr Bhaskaran.  

Banks are so focused on choosing the right FRTB model and passing the desk-level tests that they forget one important detail, says James Phillips, global head of regulatory strategy at Lombard Risk. “Once those numbers are done, they have got to report them. That impact is overlooked,” he says. 

Banks need to gather data at raw level, clean the information, compute it and aggregate the results to present to their regulator. Lombard Risk works with financial institutions to help with the regulatory reporting requirements.

The huge range of financial regulation facing banks are largely driven by data, as regulators push to keep a close eye on what banks are getting up to.

“We see a complete overhaul of how banks gather, organise and share data across different functions, and how they consume data. Firms are looking to redesign what their whole end-to-end world looks like,” says Mr Phillips.

Political slowdown

FRTB has been in the pipeline since 2012 and banks are well underway in their impact assessments, says Ms Raufuss.

They are using 2017 and the first half of 2018 to implement new risk models, the rest of 2018 and the first half of 2019 to complete dry runs and will seek regulatory approval by the end of 2019.

“This is the expected timeline everyone is working against,” she says.

But earlier this year, the vice-chairman of the Financial Services Committee in the US House of Representative wrote to Federal Reserve chief Janet Yellen to warn her that she should re-evaluate the US’s commitment to international rules from the BCBS. Congressman Patrick McHenry described agreements like the Basel III accords as the result of an “opaque decision-making process” that “unfairly penalized the American financial system”.  

He wrote: “The international standards were then turned into domestic regulations that forced American firms of various sizes to substantially raise their capital requirements, leading to slower economic growth here in America.”

He called on the Federal Reserve to cease all attempts to negotiate binding standards until “President Trump has had an opportunity to nominate and appoint officials that prioritize America’s best interests”. Ms Yellen’s term as chair comes to an end in February 2018, but it is anyone’s guess whether she will be reappointed or replaced and what impact that will have on the Fed’s guidance on BCBS rules for US banks.

Banks sat up and took notice of this development, says Mr Wilcockson. MathWorks has already seen some banks reduce their focus on the Basel-inspired FRTB, as they speculate that already-delayed timelines will extend further, and may continue to soften its implementation.

“Regulatory ambivalence too perhaps encourages internal de-prioritisation, as the cost-benefit of moving to modellable from non-modellable risk factors may be a little less obvious and a little more in the future than before,” he says.

Even in the European Union, which is ahead of the pack, FRTB will not be fully implemented until at least 2025, say market experts. In December, the European Commission published the draft European legislation that will implement FRTB, which stated that the proposed amendments will start entering into force in 2019 “at the earliest”.

The fine detail in the ensuing text says that the date of application of FRTB is two years from 2019, and there is an additional three-year phase-in period during which regulators give banks a 65% discount factor.

“The EC is proposing a carefully considered progressive implementation of FRTB,” says Mr Phillips. “It would be getting on for 2025 before full FRTB implementation in Europe.”